Research On The Component Classification And Retrieval Technologies In Multi-Library

Software reuse is a development solution to avoid duplication of work in software, effective software reuse is a practical way to improve software productivity and quality. However, as a infrastructure of software reuse, component library plays an important role in building reusage in software development process. The research on component library is drawing more widely and in-depth attention of domestic and international academic. With the evolvement of technology and the field of component library, a lot of complex components and different types of component libraries emerge. So how to achieve efficient, accurate and comprehensive retrieving in multi-component library is gradually becoming a hot spot.The description and classification of software component, which is the basis for component retrieval, is related to the retrieval efficiency, understandable level and maintainability of component. In practice, the chaotic component is divided into several areas according to different standards. The development of reusable software components, starting from the original code-level components to other products which are used in software development process such as requirements analysis-level components, design-level component, test cases-level components, signifies that the technology of software reuse has gone deep into all areas of the software industry. At the same time, users' requirements for the needs and quality of reusable components are also increasing. As components are nowadays characterized by large quantity, miscellaneous function, fast change, various needs of users and so on, this paper studies of multi-component library retrieval technology, describes the characteristics of components' models which are based on keywords, facets, enumerated classification, and compares the advantages and disadvantages of the various classifications retrieval technology. It proposes a component retrieving in multi-library which is based on retrieval conditions transformation and feedback by establishing different components libraries with multiple classification.The paper starts from the technical background and the present research situation of the multi-library retrieval technology, and analysises of the various components classification retrieval method, then proposes a algorithm of retrieval conditions transformation in multi-library. This algorithm puts the retrieve information entered by the user converted into which can be recognized by the various component libraries. It achieves resource sharing between the different classification component libraries. At the same time, this paper introduces the concept of domain ontology, that if the users are not satisfied with search results, the system will carry out the secondary semantic-based retrieval, and with the ideological relevance feedback and clustering, and repeated interaction with users' queries, it will retrieve a series of components in semantically related with users' interestes and provide users with the evaluation criteria, which helps users to choose components for their own needs and makes the retrieval more in line with users' intention. Through domain knowledge in domain ontology, this paper makes the users'retrieval information refine semantically and expand grammatically, so as to effectively improve retrieval recall and precision of the component retrieval.Finally, a multi-library component retrieval system is constructed to compare and analyse this retrieval method. The results indicate that two algorithms proposed in this paper have better retrieval efficiency and are flexible, effective, theoretically significant and practically valuable.